Elite sports are about passion and unpredictable results. Coaches make decisions based on their gut and influence games. It's the way it has always been — up until very recently. Nowartificial intelligence (AI) is starting to influence coaching at elite level. But will its use make sports more predictable and less emotional?
Data has been used to analyze games for years, but when it comes to evaluation, coaches are increasingly reaching the limits of their capacity. That's because game data is becoming more accurate and extensive. This is the field where AI should help in the future.
"Artificial intelligence is a technology executed by computers and machines to perform tasks that normally require human intelligence but are usually too complex for people," says sports scientist Daniel Memmert from the German Sport University Cologne.
"These AI systems use data and algorithms to autonomously recognize and objectively evaluate patterns of play."
It is precisely this objective evaluation that is distinct from the often subjective assessments of a coach, adds Memmert.
Will AI replace humans?
AI has revolutionized data-based analysis and is having a greater influence on tactical decisions, injury prevention and training load.
"Sports clubs have brutally expanded the use of AI," Memmert explains. "Clubs have realized how much they can do with the data they collect."
Sport clubs are also expanding their expertise. Rather than employing people with experience in a specific sport, clubs are turning more and more to programmers and computer scientists.
"In 2017, Barcelona employed 15 new data and video analysts, as well as data scientists,” Memmert says, adding: "The use of AI systems will lead to more people working in elite sport, not fewer.”
Because, as Memmert explains, teams will always need experts who know the sport really well and who can work with the information and metrics that AI delivers.
Schalke scouting with the help of AI
In the meantime, AI is being used in a variety of areas. At Germany's Bundesliga 2 club Schalke, scouts are using AI to support their search for new players.
"We have defined about 40 values in our scouting and for three of them we use artificial intelligence and decide whether these are values that make the difference for us," Schalke sports director Peter Knäbel told German public broadcaster ZDF.
"For example, you define a player's change-of-space behavior, for example, the accuracy with which he passes the ball. And put artificial intelligence on top of that."
In this way, it is easy to see whether a footballer has a similar value as a player who has departed. Nevertheless, the final decision about a signing still lies with the scout and the coaches. In this case, AI is another way to better assess potential Schalke players ahead of time.
'AI is like a small child'
Intelligent systems are also being used in handball. For years, players in the Handball Bundesliga have worn sensors that collect millions of pieces of information, including distance covered, speed, positional play and biometric data.
"We feed the computer with this information so it learns how a handball game works,” says coach Michael Döring. "The system is, at the beginning, like a small child, and it learns how to understand the game better with the data it gets fed."
In the last two years, AI has been gathered from around 600 handball games. SG Flensburg-Handewitt's system has accumulated a great deal of knowledge about the sport and in the club's data analysis, AI can now predict the likelihood of a goal.
That means Döring and his colleagues can see which players in which positions have the best chance to score.
"Where should the line player be? How should we start the second wave? These are then suddenly things where we have added value," the sports scientist tells DW. "It's like an additional advisor. And we can use this data-based approach to make our system more successful."
Döring, who used to be a handball coach, works in Flensburg's team of experts as a facilitator between the data and the coach. Because not all the information that is fed into AI is helpful for training or the tactical approach in a game.
"We take the questions from practice and insert them into theory. And the knowledge we then get from the AI-based analysis we have to package in such a way that you can also do something with it in practice."
In the future, handball players hope that the use of artificial intelligence will help them with tactical questions but also with putting together a team that will have the highest possible probability of success.
The more complex, the more logical AI becomes
Nevertheless, the use of AI-based data remains in its infancy. The current system is only as intelligent as the information it receives from people. However, it will continue to develop rapidly, further digitalize and change sport.
"The more players there are on a field, the more complex the patterns that occur. So the more people are active in a sport, the larger amounts of data that can be collected and the more it makes sense to have this data processed by smart algorithms," says Memmert.